The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/79e3f70a5d994598ba01bba48ef622d02cdf2938

commit 79e3f70a5d994598ba01bba48ef622d02cdf2938
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Fri Apr 22 17:43:24 2016

media-libs/mesa: Removed code that causes Mesa to return a desktop GL enum

When the draw_buffer_enum_to_bitmask function
is called while Mesa is in it's meta state,
the wrong GLbitfield is returned sometimes for
OpenGL ES contexts.  This is because the
_mesa_meta_begin function makes the ctx->API always
equal to desktop GL during the meta state.

In the case of GL_BACK, the _mesa_is_gles(ctx)
function is checked when returning the enum.
This check evaluates to false during the meta state,
even on mobile devices.

BUG=b/27656575, chromium:606472
TEST=Ran dEQP-GLES2.functional.clipping.line.line_attrib_clip
with the pbuffer option with on Cyan.  Verified it
fails without the patch.

Change-Id: I111fc61dfeb1f8289ab71227c9ac9b7e6612c002
Reviewed-on: https://chromium-review.googlesource.com/340347
Commit-Ready: Ilja Friedel <ihf@chromium.org>
Tested-by: Ilja Friedel <ihf@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

[modify] https://crrev.com/79e3f70a5d994598ba01bba48ef622d02cdf2938/media-libs/mesa/mesa-11.3.0.ebuild
[add] https://crrev.com/79e3f70a5d994598ba01bba48ef622d02cdf2938/media-libs/mesa/files/11.5-meta-state-fix.patch
[rename] https://crrev.com/79e3f70a5d994598ba01bba48ef622d02cdf2938/media-libs/mesa/mesa-11.3.0-r10.ebuild

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/bff917791f92050f0ef76c393ce1adfec5ba7c46

commit bff917791f92050f0ef76c393ce1adfec5ba7c46
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Thu Apr 21 22:39:25 2016

Added support for running dEQP with --surface_type=pbuffer option

BUG=b/27656575, chromium:606472
TEST=Ran GLES2, GLES3, GLES31 on veryon-minnie. No GLES2
failures, less than 10 GLES3 and GLES31 tests now fail.

Change-Id: I5336a43a562c2bb6c53a7218bd52909870326f21
Reviewed-on: https://chromium-review.googlesource.com/340292
Commit-Ready: Ilja Friedel <ihf@chromium.org>
Tested-by: Ilja Friedel <ihf@chromium.org>
Reviewed-by: Ilja Friedel <ihf@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

[rename] https://crrev.com/bff917791f92050f0ef76c393ce1adfec5ba7c46/media-gfx/deqp/files/0002-Delete-compiler-check.patch
[rename] https://crrev.com/bff917791f92050f0ef76c393ce1adfec5ba7c46/media-gfx/deqp/files/0003-Fix-clang-signed-shift-warning.patch
[rename] https://crrev.com/bff917791f92050f0ef76c393ce1adfec5ba7c46/media-gfx/deqp/deqp-2016.03.08-r3.ebuild
[add] https://crrev.com/bff917791f92050f0ef76c393ce1adfec5ba7c46/media-gfx/deqp/files/0004-Added-support-for-creating-pBuffer-target.patch

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/2bd88d086688cf3c867bd6b300c3d42e1dfa795a

commit 2bd88d086688cf3c867bd6b300c3d42e1dfa795a
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Thu Apr 21 19:37:39 2016

Added pbuffer hooks in Mesa

This change enables the creation of pbuffer
surfaces on the surfaceless platform.

BUG=b/27656575, chromium:606472
TEST=Ran dEQP GLES2, GLES3, and GLES31 tests on cyan and alex.
Only ~100 tests should fail for GLES3 and GLES31 now on cyan.
 ~1000 GLES2 tests should fail on alex (GLES31/GLES3 tests are skipped).

Change-Id: Iba6b3eae599e07d7471c067c6f5de2b47b33ddfd
Reviewed-on: https://chromium-review.googlesource.com/340291
Commit-Ready: Ilja Friedel <ihf@chromium.org>
Tested-by: Ilja Friedel <ihf@chromium.org>
Reviewed-by: Ilja Friedel <ihf@chromium.org>

[modify] https://crrev.com/2bd88d086688cf3c867bd6b300c3d42e1dfa795a/media-libs/mesa/mesa-11.3.0.ebuild
[add] https://crrev.com/2bd88d086688cf3c867bd6b300c3d42e1dfa795a/media-libs/mesa/files/11.4-pbuffer-surfaceless-hooks.patch
[rename] https://crrev.com/2bd88d086688cf3c867bd6b300c3d42e1dfa795a/media-libs/mesa/mesa-11.3.0-r9.ebuild

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/third_party/autotest/+/42c5d9dc7ddb0fca45bc9d1801246fbaf99fcc12

commit 42c5d9dc7ddb0fca45bc9d1801246fbaf99fcc12
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Mon Apr 25 16:30:27 2016

graphics_dEQP: Run with pbuffer option on all boards
except Rogue.

There are still some changes needed to enable pbuffers
on Rogue (partner bug 52663). Until then, special case
the board.

BUG= chromium:606472 
TEST=Ran dEQP on Cyan and Oak, and verified the desired
options were present.

Change-Id: Id22c3e3981182ec71fc75f00086583a0d6e5a8ca
Reviewed-on: https://chromium-review.googlesource.com/340495
Commit-Ready: Ilja Friedel <ihf@chromium.org>
Tested-by: Ilja Friedel <ihf@chromium.org>
Reviewed-by: Ilja Friedel <ihf@chromium.org>

[modify] https://crrev.com/42c5d9dc7ddb0fca45bc9d1801246fbaf99fcc12/client/site_tests/graphics_dEQP/graphics_dEQP.py

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/642f555c6a372c5485bbcbc8e1631391dd6a8ac7

commit 642f555c6a372c5485bbcbc8e1631391dd6a8ac7
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Mon May 02 17:23:05 2016

media-libs/mesa: Enabled double buffered pbuffers on ChromeOS

Some dEQP tests rely on a double-buffered
configuration to work correctly.

BUG=b/27656575, chromium:606472
TEST=Ran dEQP GLES2, GLES3, GLES31 on Cyan. An
additional 74 GLES3 tests now pass.  No regressions.

Change-Id: I841538aa014e43c095cec2447a07a2591a3ba788
Reviewed-on: https://chromium-review.googlesource.com/342000
Commit-Ready: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

[add] https://crrev.com/642f555c6a372c5485bbcbc8e1631391dd6a8ac7/media-libs/mesa/files/11.7-double-buffered.patch
[modify] https://crrev.com/642f555c6a372c5485bbcbc8e1631391dd6a8ac7/media-libs/mesa/mesa-11.3.0.ebuild
[modify] https://crrev.com/642f555c6a372c5485bbcbc8e1631391dd6a8ac7/media-libs/mesa/mesa-9999.ebuild
[rename] https://crrev.com/642f555c6a372c5485bbcbc8e1631391dd6a8ac7/media-libs/mesa/mesa-11.3.0-r13.ebuild

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/005bd7baf8c92f373772234badee6901553d8032

commit 005bd7baf8c92f373772234badee6901553d8032
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Fri Jul 08 01:26:29 2016

media-libs/waffle: Set correct surface type

ChromeOS doesn't have native windows, Waffle assumes it does. Hardcode
the surface type here so we can drop a hack currently in Mesa (Waffle
hacks are easier to maintain!).  A more correct way to fix this would
be to pass in the surface type to the wflinfo command line, but that
would require changes in every test that uses wflinfo.

BUG= chromium:606472 
TEST=Ran:

wflinfo -p null -a gles2

Verified it worked on cyan (with Mesa hack removed) and veyron_minnie.

Change-Id: I0d74358d88802699d588b7db94b0d0f3ffc7d053
Reviewed-on: https://chromium-review.googlesource.com/358936
Commit-Ready: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: Gurchetan Singh <gurchetansingh@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

[rename] https://crrev.com/005bd7baf8c92f373772234badee6901553d8032/media-libs/waffle/waffle-1.6.0-r7.ebuild
[add] https://crrev.com/005bd7baf8c92f373772234badee6901553d8032/media-libs/waffle/files/0003-eglChooseConfig.patch